Take Home Exercise 3

Uncover illegal, unreported, and unregulated (IUU) fishing activities through visual analytics

Author

Fangxian

Published

June 18, 2023

Modified

June 18, 2023

Task

Using the data provide by the VAST challenge, we are looking into the Mini-Challenge 3 (MC3) to identify compaines possibly engaged in illegal, unreported, and unregulated (IUU) fishing.

Data Wraggling

Load Packages

Show the code

pacman::p_load(jsonlite, tidygraph, ggraph, visNetwork, graphlayouts, ggforce,skimr,tidytext, tidyverse,igraph, topicmodels,tm, stringr, ldatuning)

Data Import

In the code chunk below, fromJSON() of jsonlite package is used to import MC3.json into R environment.

Show the code

mc3_data <- fromJSON("data/MC3.json")

Examine the data, this is not a directed graph, not looking into in- and out-degree of the nodes.

Extracting edges

Below code chunk changes the links field into character field.

Show the code

mc3_edges <- as_tibble(mc3_data$links)%>%
  distinct() %>%
  mutate(source = as.character(source),
         target = as.character(target),
         type = as.character(type)) %>%
  group_by(source, target, type) %>%
    summarise(weights = n()) %>%
  filter(source!=target)%>%
  ungroup

Extracting nodes

Show the code

mc3_nodes <- as_tibble(mc3_data$nodes) %>%
#  distinct()%>%
  mutate(country = as.character(country),
         id = as.character(id),
         product_services = as.character(product_services),
         revenue_omu = as.numeric(as.character(revenue_omu)),
         type = as.character(type)) %>%
    select(id, country, type, revenue_omu, product_services)

Initial Data Exploration

Exploring the edges dataframe

In the code chunk below, skim() of skimr package is used to display the summary statistics of mc3_edges tibble data frame.

Show the code

skim(mc3_edges)

Data summary
Name	mc3_edges
Number of rows	24036
Number of columns	4
_______________________
Column type frequency:
character	3
numeric	1
________________________
Group variables	None

Variable type: character

skim_variable	complete_rate	min	max	n_unique
source	1	6	700	12856
target	1	6	28	21265
type	1	16	16	2

Variable type: numeric

skim_variable	n_missing	complete_rate	mean	sd	p0	p25	p50	p75	p100	hist
weights	0	1	1	0	1	1	1	1	1	▁▁▇▁▁

The report above reveals that there is not missing values in all fields.

In the code chunk below, datatable() of DT package is used to display mc3_edges tibble data frame as an interactive table on the html document.

Show the code

DT::datatable(mc3_edges)

Below code chunks, counting number of companies a person owns.

Show the code

ggplot(data = mc3_edges,
       aes(x=type)) +
  geom_bar()

Show the code

unique_ids <- unique(mc3_edges$target)
num_unique_ids <- length(unique_ids)
num_unique_ids

[1] 21265

Show the code

Noofcompanies <- mc3_edges %>%
  group_by(target, source, type) %>%
  filter(type == "Beneficial Owner") %>%
  summarise(count=n()) %>%
  group_by(target)%>%
  summarise(count=sum(count))

psych::describe(Noofcompanies)

        vars     n   mean      sd median trimmed     mad min   max range skew
target*    1 15305 7653.0 4418.32   7653    7653 5672.43   1 15305 15304 0.00
count      2 15305    1.1    0.40      1       1    0.00   1     9     8 6.28
        kurtosis    se
target*    -1.20 35.71
count      61.69  0.00

Below code chunk summarizes the numbers of owners to the company.

Show the code

Noofowners <- mc3_edges %>%
  group_by(source, target, type) %>%
  summarise(count=n()) %>%
  group_by(source)%>%
  summarise(count=sum(count))

psych::describe(Noofowners)

        vars     n    mean      sd median trimmed     mad min   max range  skew
source*    1 12856 6428.50 3711.35 6428.5 6428.50 4765.08   1 12856 12855  0.00
count      2 12856    1.87    3.47    1.0    1.22    0.00   1   120   119 11.36
        kurtosis    se
source*    -1.20 32.73
count     215.82  0.03

Below code chunk we are interested to see top 50 owners owning multiple companies, with John Smith and Michael Johnson have the highest of 9 companies to their name. This could be suspicious as why they need so many companies.

Show the code

list_top_50 <- Noofcompanies %>%
  arrange(desc(count)) %>%
  top_n(50, wt = count) 

ggplot(data = list_top_50, 
       aes(x = reorder(target, -count), y = count)) +
  geom_bar(stat = "identity") +
  theme(axis.text.x = element_text(angle = 90, hjust = 1))

Exploring the nodes dataframe

Show the code

skim(mc3_nodes)

Data summary
Name	mc3_nodes
Number of rows	27622
Number of columns	5
_______________________
Column type frequency:
character	4
numeric	1
________________________
Group variables	None

Variable type: character

skim_variable	complete_rate	min	max	n_unique
id	1	6	64	22929
country	1	2	15	100
type	1	7	16	3
product_services	1	4	1737	3244

Variable type: numeric

skim_variable	n_missing	complete_rate	mean	sd	p0	p25	p50	p75	p100	hist
revenue_omu	21515	0.22	1822155	18184433	3652.23	7676.36	16210.68	48327.66	310612303	▇▁▁▁▁

The report above reveals that there is no missing values in all fields.

In the code chunk below, datatable() of DT package is used to display mc3_nodes tibble data frame as an interactive table on the html document.

Show the code

DT::datatable(mc3_nodes)

Below code chunk to find out how is the distribution among the types of ownerhships.

Show the code

ggplot(data = mc3_nodes,
       aes(x = type)) +
  geom_bar()

Below code chunk we check on the revenue distribution among the types of ownerships.

Show the code

ggplot(data = mc3_nodes,
       aes(x= type,
         y = revenue_omu)) +
  geom_boxplot()

We combined the nodes and edges data so we can find out more on the owner-company relationships.

Show the code

combined <- left_join(mc3_nodes,mc3_edges,
                  by=c("id"="source"))

Below code chunk to find out more on which owners have high number of companies also generating a lot of revenue.

Show the code

combined <- combined %>%
  group_by(target, type.y, id, country, type.x, product_services)%>%
  summarize(revenue_omu) %>%
  filter(type.y == "Beneficial Owner")

filtered_combined <- combined %>%
  filter(target %in% list_top_50$target)%>%
  arrange(desc(revenue_omu))

ggplot(data = filtered_combined, 
       aes(x = target, y = revenue_omu)) +
  geom_bar(stat = "identity") +
  theme(axis.text.x = element_text(angle = 90, hjust = 1))

Michael Johnson, Mark Miller and James Rodriguez stand out from the above chart, below code we want to see what business they did that generate more revenue.

Show the code

Top_3_Revenue<- combined %>%
  filter (target %in% c("Michael Johnson", "Mark Miller","James Rodriguez")) %>%
  arrange(desc(revenue_omu))

DT::datatable(Top_3_Revenue)

Insights

From the above data table, we see that Michael Johnson is involved in the fishing business and having many companies in different countries. The FishEye could probably look more into his business landscape across different companies and his business activities to understand more.

Text Sensing with tidytext

Simple word count

The code chunk below calculates number of times the word fish appeared in the field product_services.

Show the code

mc3_nodes %>% 
    mutate(n_fish = str_count(product_services, "fish"))

# A tibble: 27,622 × 6
   id                          country type  revenue_omu product_services n_fish
   <chr>                       <chr>   <chr>       <dbl> <chr>             <int>
 1 Jones LLC                   ZH      Comp…  310612303. Automobiles           0
 2 Coleman, Hall and Lopez     ZH      Comp…  162734684. Passenger cars,…      0
 3 Aqua Advancements Sashimi … Oceanus Comp…  115004667. Holding firm wh…      0
 4 Makumba Ltd. Liability Co   Utopor… Comp…   90986413. Car service, ca…      0
 5 Taylor, Taylor and Farrell  ZH      Comp…   81466667. Fully electric …      0
 6 Harmon, Edwards and Bates   ZH      Comp…   75070435. Discount superm…      0
 7 Punjab s Marine conservati… Riodel… Comp…   72167572. Beef, pork, chi…      0
 8 Assam   Limited Liability … Utopor… Comp…   72162317. Power and Gas s…      0
 9 Ianira Starfish Sagl Import Rio Is… Comp…   68832979. Light commercia…      0
10 Moran, Lewis and Jimenez    ZH      Comp…   65592906. Automobiles, tr…      0
# ℹ 27,612 more rows

Tokenisation

The word tokenisation have different meaning in different scientific domains. In text sensing, tokenisation is the process of breaking up a given text into units called tokens. Tokens can be individual words, phrases or even whole sentences. In the process of tokenisation, some characters like punctuation marks may be discarded. The tokens usually become the input for the processes like parsing and text mining.

In the code chunk below, unnest_token() of tidytext is used to split text in product_services field into words.

Show the code

token_nodes <- mc3_nodes %>%
  unnest_tokens(word, 
                product_services)

The two basic arguments to unnest_tokens() used here are column names. First we have the output column name that will be created as the text is unnested into it (word, in this case), and then the input column that the text comes from (product_services, in this case).

Show the code

token_nodes %>%
  count(word, sort = TRUE) %>%
  top_n(15) %>%
  mutate(word = reorder(word, n)) %>%
  ggplot(aes(x = word, y = n)) +
  geom_col() +
  xlab(NULL) +
  coord_flip() +
      labs(x = "Count",
      y = "Unique words",
      title = "Count of unique words found in product_services field")

The bar chart reveals that the unique words contains some words that may not be useful to use. For instance “a” and “to”. In the word of text mining we call those words stop words. You want to remove these words from your analysis as they are fillers used to compose a sentence.

Using filter we also discover many “character(0)” which has no meaning in itself, we will also proceed to replace them with “NA”.

Removing stopwords

Show the code

token_nodes$word[token_nodes$word == "character"] <- "NA"
token_nodes$word[token_nodes$word == "0"] <- "NA"

Show the code

stopwords_removed <- token_nodes %>% 
  anti_join(stop_words)

Show the code

stopwords_removed %>%
  count(word, sort = TRUE) %>%
  top_n(15) %>%
  mutate(word = reorder(word, n)) %>%
  ggplot(aes(x = word, y = n)) +
  geom_col() +
  xlab(NULL) +
  coord_flip() +
      labs(x = "Count",
      y = "Unique words",
      title = "Count of unique words found in product_services field")

Below code removed the top 3 irrelevant words.

Show the code

stopwords_removed %>%
  count(word, sort = TRUE) %>%
  top_n(20) %>%
  mutate(word = reorder(word, n)) %>%
  filter(!word %in% head(word, 3)) %>%
  ggplot(aes(x = word, y = n)) +
  geom_col() +
  xlab(NULL) +
  coord_flip() +
  labs(x = "Count",
       y = "Unique words",
       title = "Count of unique words found in product_services field")

Topic Modelling

This section, we will conduct topic modelling to find out the similar business groups. For this, we will make use of Latent Dirichlet Allocation (LDA).

First, run the below code chunk to remove the irrelevant top 3 keywords: “NA”, “unknown”, “products”.

Show the code

stopwords_removed <- stopwords_removed %>%
  filter(!word %in% c("NA", "unknown", "products"))

dim(stopwords_removed)

[1] 44082     5

Converting Document-Term Matrix

We need to convert the document to a document-term matrix to to run LDA.

Show the code

# using as.matrix()
dtm <- stopwords_removed %>%
  count(id, word) %>%  # count each word used in each identified review 
  cast_dtm(id, word, n) %>%  # use the word counts by reviews  to create a DTM
  as.matrix()

Show the code

# create models with different number of topics
result <- ldatuning::FindTopicsNumber(
  dtm,
  topics = seq(from = 2, to = 20, by = 1),
  metrics = c("Griffiths2004", "CaoJuan2009", "Arun2010", "Deveaud2014"),
  method = "Gibbs",
  control = list(seed = 77),
  verbose = TRUE
)

fit models... done.
calculate metrics:
  Griffiths2004... done.
  CaoJuan2009... done.
  Arun2010... done.
  Deveaud2014... done.

Finding optimal number of topics

Show the code

FindTopicsNumber_plot(result)

From the above graph, we see that after topic 12, for “Griffiths2004”, there isn’t much increase in the value, hence, we choose topics to be 12.

Show the code

# number of topics
K <- 12
# set random number generator seed
set.seed(1234)
# compute the LDA model, inference via 1000 iterations of Gibbs sampling
topicModel <- LDA(dtm, K, method="Gibbs", control=list(iter = 500, verbose = 25))

K = 12; V = 7746; M = 3894
Sampling 500 iterations!
Iteration 25 ...
Iteration 50 ...
Iteration 75 ...
Iteration 100 ...
Iteration 125 ...
Iteration 150 ...
Iteration 175 ...
Iteration 200 ...
Iteration 225 ...
Iteration 250 ...
Iteration 275 ...
Iteration 300 ...
Iteration 325 ...
Iteration 350 ...
Iteration 375 ...
Iteration 400 ...
Iteration 425 ...
Iteration 450 ...
Iteration 475 ...
Iteration 500 ...
Gibbs sampling completed!

Show the code

lda_topics <- topicModel %>%
  tidy(matrix = "beta")

Show the code

lda_topics <- LDA(
  dtm,
  k = 12,
  method = "Gibbs",
  control = list(seed=42)
  ) %>%
  tidy(matrix = "beta")
word_probs <- lda_topics %>%
  group_by(topic) %>%
  top_n(15, beta) %>%
  ungroup() %>%
  mutate(term2 = fct_reorder(term, beta))
ggplot(
  word_probs,
  aes(term2, beta, fill=as.factor(topic))
  ) +
  geom_col(show.legend = FALSE) +
  facet_wrap(~ topic, scales = "free") +
  coord_flip()

Insights

From the above result, we can see topic 5 and topic 10 are most related to fishing industry. Topic 5 is a group of fishing business likely to have fresh sea products, while topic 10 is a group of fishing business likely to have more processed sea products. In the following network analysis session, we will focus on the nodes that contains topic 5 and topic 10 keywords.

Initial Network Visualization and Analysis

Building network model with tidygraph

From the above text insights, we are interested to see the network of companies of Beneficial Owners with topic 5 as their product services.

Show the code

mc3_nodes_topic5 <- stopwords_removed %>%
  filter(stopwords_removed$word %in% c("fish","salmon","tuna","shrimp","squid","frozen","sea","cod","fillets","crabs","fillet","pollock","lobster","smoked"))

mc3_nodes_topic10 <- stopwords_removed %>%
  filter(stopwords_removed$word %in% c("seafood","fish","fresh","processing","seafoods","marine","shellfish","frozen","aquatic","oils","canning","packing","fats","cured","preparation"))

Show the code

mc3_edges_topic5 <- mc3_edges[mc3_edges$source %in% mc3_nodes_topic5$id,] %>%   
  filter(type == "Beneficial Owner")  
id1 <- mc3_edges_topic5 %>%   
  select(source) %>%   
  rename(id = source) 
id2 <- mc3_edges_topic5 %>%    
  select(target) %>%    
  rename(id = target)  
mc3_nodes_topic5 <- rbind(id1, id2) %>%   
  distinct() %>%    
  left_join(mc3_nodes_topic5,             
            unmatched = "drop") 


mc3_edges_topic10 <- mc3_edges[mc3_edges$source %in% mc3_nodes_topic10$id,] %>%   
  filter(type == "Beneficial Owner")  
id1 <- mc3_edges_topic10 %>%   
  select(source) %>%   
  rename(id = source) 
id2 <- mc3_edges_topic10 %>%    
  select(target) %>%    
  rename(id = target)  
mc3_nodes_topic10 <- rbind(id1, id2) %>%   
  distinct() %>%    
  left_join(mc3_nodes_topic10,             
            unmatched = "drop")

Show the code

mc3_graph_topic5 <- tbl_graph(nodes = mc3_nodes_topic5,                                               edges = mc3_edges_topic5,                                                 directed = FALSE)    
mc3_graph_topic5 <-mc3_graph_topic5 %>%   
  mutate(betweenness=centrality_betweenness())  

mc3_graph_topic10 <- tbl_graph(nodes = mc3_nodes_topic10,                                               edges = mc3_edges_topic10,                                                 directed = FALSE)    
mc3_graph_topic10 <-mc3_graph_topic10 %>%   
  mutate(betweenness=centrality_betweenness())

Using the distribution function to understand the centrality_betweenness() for topic 5 and 10 graphs.

Show the code

ggplot(as.data.frame(mc3_graph_topic5),aes(x=betweenness))+
  geom_histogram(bins=10,fill="lightblue",colour="black")+
  ggtitle("Distribution of centrality betweenness for topic 5")+
  theme(plot.title = element_text(hjust=0.5))

Show the code

ggplot(as.data.frame(mc3_graph_topic10),aes(x=betweenness))+
  geom_histogram(bins=10,fill="lightblue",colour="black")+
  ggtitle("Distribution of centrality betweenness for topic 10")+
  theme(plot.title = element_text(hjust=0.5))

Looking at this, we can filter our records where the centrality between is greater than 75 to understand the interactions.

Show the code

set.seed (1234)
degrees <- degree(mc3_graph_topic5)
V(mc3_graph_topic5)$degree <- degrees

mc3_graph_topic5 %>%
  filter(betweenness >= 75) %>%
ggraph(layout = "fr") +
  geom_edge_link(aes(alpha=0.5)) +
  geom_node_point(aes(
    size = betweenness,
    colors = "lightblue",
    alpha = 0.5)) +
  scale_size_continuous(range=c(1,10))+
  geom_node_text(aes(label = id, filter= betweenness >=200 &degree >0), repel = TRUE)+
  theme_graph()

Show the code

set.seed (1234)
degrees <- degree(mc3_graph_topic10)
V(mc3_graph_topic10)$degree <- degrees

mc3_graph_topic10 %>%
  filter(betweenness >= 75) %>%
ggraph(layout = "fr") +
  geom_edge_link(aes(alpha=0.5)) +
  geom_node_point(aes(
    size = betweenness,
    colors = "lightblue",
    alpha = 0.5)) +
  scale_size_continuous(range=c(1,10))+
  geom_node_text(aes(label = id, filter= betweenness >=200 &degree >0), repel = TRUE)+
  theme_graph()

Insights and future work

Looking at the above two network graph, we see there are some overlapping companies in these two topic groups. Upcoming work will look into the similarities between these two groups.